Getting Started

TraceVault is the flight recorder for AI-assisted development. It captures every AI coding session -- prompts, responses, code changes, and tool invocations -- so your team always has a complete audit trail.

Quick Start

1. Install the CLI

curl -fsSL https://tracevault.dev/install.sh | sh

Or install via Cargo:

cargo install tracevault

2. Initialize a repository

In your project directory, run:

tracevault init

This creates a .tracevault/ directory with a local configuration file. It also installs Claude Code hooks — by default into .claude/settings.json (shared with the team). To install them into your personal .claude/settings.local.json instead, pass --claude-settings local. When the flag is omitted on a TTY, init prompts for the choice; in non-interactive environments it defaults to shared. See CLI reference for details.

3. Start recording

TraceVault hooks into your AI coding tools automatically. Once initialized, every session with Claude Code, Cursor, or Copilot is captured:

# Sessions are recorded automatically when your AI tool runs
# View recent sessions:
tracevault log

# View details of a specific session:
tracevault show <session-id>

4. Stream traces to a server (optional)

For team-wide visibility, log in to a TraceVault server. After tracevault init, every session and commit is streamed automatically via the installed Claude Code and git hooks — no manual upload step.

tracevault login --server-url https://your-tracevault-server.example.com
tracevault init

Key Concepts

  • Session -- A single AI coding interaction (one conversation with an AI assistant)
  • Trace -- The full record of a session: prompts, responses, tool calls, and file diffs
  • Policy -- Rules that define what must be captured and what should be redacted
  • Repository -- A project directory tracked by TraceVault

What's Next?

  • CLI Reference -- Full command reference
  • Server Setup -- Deploy a TraceVault server for your team
  • Post-Install Setup -- Configure LLM provider, GitHub integration, and team access after deploying the server
  • Policies -- Configure and enforce tool-call requirements per repository
  • API Reference -- Integrate TraceVault into your workflow